We present a new partitioning method to determinate fuzzy regions in fuzzy probabilistic predictors. Fuzzy probabilistic predictors are modifications of discrete probabilistic classifiers, as Naive Bayes Classifier and Hidden Markov Model, in order to enable them to predict continuous values. Two fuzzy probabilistic predictors, Fuzzy Markov Predictor and the Fuzzy Hidden Markov Predictor, are applied to the task of monthly electric load single-step forecasting using this new partitioning and successfully compared with two Kalman Filter Models, and two traditional forecasting methods, Box-Jenkins and Winters exponential smoothing. The employed time series present a sudden significant changing behavior at their last years, as it occurs in an energy rationing.